Contact frequency determines outcome of basal insulin initiation trials in type 2 diabetes

Aims/hypothesis The aim of the present study was to investigate whether predetermined contact frequency with the study teamand endpoint insulin dose are associated with study outcomes in basal insulin initiation trials in type 2 diabetes. Methods A systematic Medline search was performed. Using data from the selected studies, contact frequency was plotted against HbA(1c) reduction and endpoint insulin dose. The importance of face-to-face vs telephone contact was also analysed. Insulin dose was plotted against HbA(1c) reduction, hypoglycaemia rate and weight gain. To investigate non-specific study effects, the relationship between contact frequency and HbA(1c) was also assessed in dipeptidyl peptidase-4 (DPP-4) inhibitor trials. Results The reduction in HbA(1c) was highly correlated with contact frequency and endpoint insulin dose (r(2)=0.751, p<0.001 and r(2)=0.433, p=0.008, respectively). However, after adjusting for contact frequency, the relationship between insulin dose and HbA(1c) reduction was no longer significant (p=0.270). The frequency of both clinical and telephone contacts were independent predictors of HbA(1c) improvement (p=0.010 and p<0.001, respectively). We found no dose response relationship between end-of-study insulin dose and hypoglycaemia or weight gain. In DPP-4 inhibitor studies, contact frequency was not positively associated with HbA(1c). Conclusions/interpretation The frequency of contact with the study team is highly correlated with the improvement in HbA(1c) achieved in basal insulin initiation trials in type 2 diabetic patients. This has important implications for trial design and interpretation, as well as for clinical care.